V5.20
🔍

ARIA CONNECTOR EA V5

AI-Powered Expert Advisor for MetaTrader 5

Connect up to 14 AI providers, 75+ models — including 4 providers optimized for China — and 28+ professional trading systems in one powerful EA.

14 AI Providers
75+ AI Models
China-Ready
Multi-AI Voting
Risk Sentinel
Prop Firm Ready

Quick Start Guide

Get ARIA running in under 5 minutes
1. Install EA
2. Get API Key
3. Edit keys.txt
4. Enable WebRequest
5. Attach to Chart
6. Trade!
  1. Install the EA — Copy files to your MT5 Experts folder, including aria-modules.
  2. Get an API key — Go to openrouter.ai/keys (or open.bigmodel.cn for China users). OpenRouter requires a one-time $10 deposit to unlock all free models.
  3. Edit keys.txt — In MT5: File → Open Data Folder → MQL5 → Files. Paste your key in the correct line.
  4. Enable WebRequestTools → Options → Expert Advisors. Add the URL for your AI provider.
  5. Attach EA — Drag onto any chart. Enable "AutoTrading" (green button).
WebRequest URLs Reference
ProviderURL to Add
OpenRouterhttps://openrouter.ai
OpenAIhttps://api.openai.com
Claudehttps://api.anthropic.com
Geminihttps://generativelanguage.googleapis.com
DeepSeekhttps://api.deepseek.com
Mistralhttps://api.mistral.ai
Grokhttps://api.x.ai
Llamahttps://api.llama-api.com
Qwen (China)https://dashscope.aliyuncs.com
Zhipu GLM (China)https://open.bigmodel.cn
Moonshot (China)https://api.moonshot.cn
Baichuan (China)https://api.baichuan-ai.com
Groqhttps://api.groq.com
Cerebrashttps://api.cerebras.ai
News Filterhttps://nfs.faireconomy.media
Telegramhttps://api.telegram.org
★ China Users Quick Start
If you are in mainland China, we recommend Zhipu GLM with the GLM-4-Flash model — it's completely free with no deposit required (unlike OpenRouter's $10 deposit). Go to open.bigmodel.cn, create an account, generate a key, paste it in the zhipu: line of keys.txt, add the WebRequest URL, and you're ready to trade with AI. No VPN needed.
AI Configuration

AI Provider Selection

Choose from 14 AI providers including 4 China-accessible options
ParameterDefaultDescription
Select AI ProviderOpenRouter AI #1Selects which AI service will analyze the market

ARIA V5.20 supports 14 AI providers (18 connection slots total), including 4 providers specifically added for users in mainland China where Western AI services are blocked.

Global Providers
OpenRouter (5 slots) Free Available*
AI marketplace with 7 verified free and 75+ paid models. Best starting point.
OpenAI Paid
GPT-4 Turbo, GPT-5 series. Industry standard for AI analysis.
Claude (Anthropic) Paid
Superior reasoning and deep analysis. Claude Sonnet 4 recommended.
Gemini (Google) Free Available
Gemini 2.5 Pro for highly detailed market analysis.
Llama (Meta) Paid
Open-source models. Llama 405B for maximum quality.
DeepSeek Paid
Excellent quality at very low cost. DeepSeek V3 / R1.
Mistral AI Paid
Fast and efficient. Mistral Large 2.1 for speed.
Grok (xAI) Paid
Real-time X/Twitter sentiment data. Grok 4 latest.
Groq Free
Custom LPU hardware for the fastest inference speeds available. Up to ~14,400 requests/day free.
Cerebras Free
World's largest AI chip (wafer-scale engine). Sub-1-second responses. ~1,000 requests/day free.

* OpenRouter requires a one-time $10 deposit to unlock access to their free AI models. After this single payment, all free models are available with no further charges.

China-Accessible Providers
Qwen (Alibaba) China
Alibaba Cloud's flagship AI. Qwen Max for deep reasoning. Affordable.
Zhipu GLM China 100% Free
GLM-4-Flash is completely free — no deposit needed. Best free option for China users.
Moonshot / Kimi China
Massive context windows up to 128K tokens. Excellent for data-heavy analysis.
Baichuan China
Specialized in Chinese language. Baichuan 4 for top-tier quality.
★ Why China-Specific Providers?
Users in mainland China face internet restrictions that block access to OpenAI, Claude, Gemini, and other Western AI services. These 4 providers (Qwen, Zhipu, Moonshot, Baichuan) were specifically added to ARIA V5 so that Chinese users can fully utilize AI-powered trading without VPN. All 4 work natively within China and offer quality comparable to Western alternatives.

API Keys Setup

Obtain and configure credentials for all 14 providers

An API Key is a special credential from the AI provider that allows ARIA to communicate with its service.

Where to Get Your API Key
ProviderWebsiteKey Format
OpenRouteropenrouter.ai/keyssk-or-v1-xxx...
OpenAIplatform.openai.com/api-keyssk-proj-xxx...
Claudeconsole.anthropic.com/settings/keyssk-ant-api03-xxx...
Geminiaistudio.google.com/apikeyAIzaSyxxx...
DeepSeekplatform.deepseek.com/api_keyssk-xxx...
Mistralconsole.mistral.ai/api-keysVaries
Grokconsole.x.aiVaries
Groqconsole.groq.comVaries
Cerebrascloud.cerebras.aiVaries
Qwendashscope.console.aliyun.comsk-xxx...
Zhipu GLMopen.bigmodel.cnVaries
Moonshotplatform.moonshot.cnsk-xxx...
Baichuanplatform.baichuan-ai.comsk-xxx...
keys.txt File Format

Open MQL5 → Files → keys.txt with any text editor:

openai:
llama:
gemini:
claude:
deepseek:
mistral:
grok:
openrouter:
groq:
cerebras:
qwen:
zhipu:
moonshot:
baichuan:

Paste your key after the colon on the correct line. Save and restart the EA.

✖ Security
  • NEVER share your keys.txt or API keys with anyone
  • API keys grant access to your provider account and may incur charges
  • If compromised, revoke immediately at the provider's website
  • ARIA only sends your key to the selected AI provider
⦿

OpenRouter — 5 AI Slots, One Key

Access 7 verified free and 75+ paid models from a single account

OpenRouter is an AI marketplace: one account, one key, dozens of models. ARIA V5.20 gives you 5 independent slots, each configurable with a different model.

Parameter (per slot)DefaultDescription
Use Free modelsEnabledToggle free vs. paid models
Free model selectionVariesChoose from 7 verified free models
Paid model selectionVariesChoose from 75+ paid models
⚠ Important Changes to Free Models (v5.20)
After extensive testing with our new ARIA Model Tester, we identified significant instability issues with OpenRouter's free models. Out of 29 free models available, only 7 return correctly formatted responses that ARIA can interpret. Many models were discontinued, return incomplete responses (missing ADVICE, CONFIDENCE, SL or TP), or are frequently rate-limited (HTTP 429). Only models scoring 5/5 on our quality test remain in the list.
7 Verified Free Models (sorted by response time)
#ModelAvg. Response Time
1google/gemma-3n-e4b-it:free5.6s
2arcee-ai/trinity-mini:free6.3s
3google/gemma-3n-e2b-it:free6.5s
4openai/gpt-oss-20b:free8.8s
5openai/gpt-oss-120b:free10.1s
6google/gemma-3-12b-it:free11.2s
7arcee-ai/trinity-large-preview:free50.2s
ℹ Note
Free models on OpenRouter can change, be removed, or become rate-limited at any time without notice. If you experience errors, try switching to a different model from the dropdown. For production trading, we recommend using paid models or the new free providers (Groq, Cerebras, Gemini) detailed below.
⚠ One-Time Deposit
OpenRouter requires a one-time $10 deposit to unlock access to all their free AI models. This is a single payment — you pay once and never again for free models. After this deposit, all free models are available with no recurring charges.

New Free AI Providers — Groq, Cerebras & Gemini

Direct API connections — no OpenRouter needed. Fast, reliable, and free.
✔ New in v5.20
We added two new free AI providers that connect directly to their APIs — no OpenRouter intermediary needed. These providers offer fast, reliable, and high-quality responses for trading analysis.
🟢 Groq (FREE)

Groq uses custom LPU (Language Processing Unit) hardware, delivering the fastest inference speeds available — typically faster than any other provider.

ModelDescription
Llama 3.3 70B VersatileBest all-around model, versatile and fast
Llama 3.1 8B InstantUltra-fast, lightweight model
Llama 4 Scout 17BLatest Llama model, good reasoning
Qwen3 32BStrong analytical model
GPT-OSS 120BLarge model, detailed analysis
Kimi K2 InstructHigh-quality instruction following
Rate Limit
30 req/min
Requests per minute
Token Limit
~15K/min
Tokens per minute
Daily Limit
~14,400
Requests per day

How to get your Groq API Key:

  1. Go to console.groq.com
  2. Sign up with Google, GitHub, or email
  3. Go to API Keys in the left sidebar
  4. Click Create API Key
  5. Copy the key and paste it in your keys.txt as: groq:your-api-key-here
🟢 Cerebras (FREE)

Cerebras runs on the world's largest AI chip (wafer-scale engine), delivering extremely fast inference — typically sub-1-second for short responses.

ModelDescription
Llama 3.1 8BFast and lightweight, great for quick signals
Qwen 3 235B A22BLarge model, excellent for detailed analysis
Rate Limit
30 req/min
Requests per minute
Token Limit
~60K/min
Tokens per minute
Daily Limit
~1,000
Requests per day

How to get your Cerebras API Key:

  1. Go to cloud.cerebras.ai
  2. Sign up with Google, GitHub, or email
  3. Go to API Keys in the dashboard
  4. Click Create API Key
  5. Copy the key and paste it in your keys.txt as: cerebras:your-api-key-here
🟢 Google Gemini (FREE)

Gemini remains one of the best free options with very generous limits and high-quality responses.

ModelDescription
Gemini 2.0 FlashFast, general purpose
Gemini 2.0 Flash-LiteFastest, lightweight
Gemini 1.5 FlashBalanced speed and quality
Gemini 3 FlashLatest generation
Rate Limit
15 req/min
Flash models
Token Limit
1M/min
Highest free tier available
Daily Limit
1,500
Requests per day

How to get your Gemini API Key:

  1. Go to aistudio.google.com
  2. Sign in with your Google account
  3. Click Get API Key in the left sidebar
  4. Click Create API Key and select a project
  5. Copy the key and paste it in the ARIA EA properties under Gemini API Key
💡 Tip
Gemini has the highest free token limit (1 million tokens per minute), making it ideal for detailed analysis prompts.
📊 Which Free Provider Should I Use?
ProviderSpeedQualityDaily LimitBest For
Groq⚡ Fastest⭐⭐⭐⭐~14,400/dayHigh-frequency queries
Cerebras⚡ Very fast⭐⭐⭐⭐~1,000/dayQuick and reliable analysis
Gemini🔄 Fast⭐⭐⭐⭐⭐1,500/dayBest quality + highest token limit
OpenRouter🔄 Variable⭐⭐⭐Model dependentVariety of models to test
💡 Recommendations
  • For best quality: Use Gemini as your primary AI
  • For fastest response: Use Groq
  • For Multi-AI mode: Combine Gemini + Groq + Cerebras + 1-2 OpenRouter models for the most robust consensus signals

China AI Providers — Qwen, Zhipu, Moonshot, Baichuan

4 providers designed for users in mainland China where Western services are blocked
★ Who Are These For?
These 4 AI providers were specifically added to ARIA V5 for users operating from mainland China, where services like OpenAI, Claude, Gemini, and other Western providers are blocked by internet restrictions. All 4 Chinese providers work natively without VPN and deliver quality comparable to Western alternatives for market analysis.

Each Chinese provider shares the same parameter structure in the EA settings:

Parameter (per provider)Description
API KeyLoaded from keys.txt (lines: qwen: zhipu: moonshot: baichuan:)
Model to useSelect the specific model version from the dropdown
Qwen — Alibaba Cloud (通义千问)

Qwen (pronounced "Chwen") is developed by Alibaba, one of China's largest tech companies. Backed by Alibaba Cloud infrastructure, it offers excellent stability, speed, and advanced reasoning capabilities. It's one of the most powerful Chinese-made AI models available today.

ModelDescriptionBest For
Qwen MaxMost powerful model, excellent reasoningDeep analysis, important trading decisions
Qwen PlusGood balance of quality and speedGeneral daily use
Qwen TurboFastest and most affordableFrequent queries, automatic mode
Qwen LongSpecialized in very long textsAnalysis with extensive historical data

How to get your Qwen API Key:

  1. Go to dashscope.console.aliyun.com
  2. Create an Alibaba Cloud account (email or Chinese phone number)
  3. Access the DashScope dashboard
  4. Navigate to API Keys and create a new key
  5. Paste the key in your keys.txt on the qwen: line
ℹ WebRequest URL
Add https://dashscope.aliyuncs.com to MT5 WebRequest settings.
Zhipu AI / GLM (智谱AI)

Zhipu AI is a Beijing-based company that develops the GLM (General Language Model) series. Their standout feature: GLM-4-Flash is completely free, making it the best option for Chinese users who want to start without any cost. The company is backed by Tsinghua University research and is one of China's leading AI labs.

ModelDescriptionBest For
GLM-4-PlusMost powerful Zhipu modelHigh-quality analysis
GLM-4Standard model, good general qualityDaily use
GLM-4-FlashFast and FREEGetting started at zero cost
GLM-4-AirLightweight and affordableFrequent, quick queries

How to get your Zhipu API Key:

  1. Go to open.bigmodel.cn
  2. Create an account (email or Chinese phone)
  3. Access the developer dashboard
  4. Navigate to API Keys and generate one
  5. Paste the key in your keys.txt on the zhipu: line
✔ Best Free Option for China
If you're in China and want to start completely free, Zhipu's GLM-4-Flash is your best bet. Unlike OpenRouter (which requires a one-time $10 deposit), Zhipu is genuinely free with no deposit required — just create an account and start using it. The only limitation is concurrency (2 simultaneous requests for new users), not daily query count. New accounts also receive a 20 million token welcome bonus to try premium models.
ℹ WebRequest URL
Add https://open.bigmodel.cn to MT5 WebRequest settings.
Moonshot / Kimi (月之暗面)

Moonshot AI is known for its popular chatbot Kimi, one of the most widely used AI assistants in China. Its key advantage is offering extremely large context windows (up to 128K tokens), allowing it to analyze significantly more market data in a single query than most other providers. This makes it particularly powerful when you enable candle history or need the AI to process large amounts of indicator data.

ModelContext WindowBest For
Moonshot V1 8K8,000 tokensQuick, affordable analysis (recommended start)
Moonshot V1 32K32,000 tokensDetailed analysis with more historical data
Moonshot V1 128K128,000 tokensExhaustive analysis with maximum data

What are tokens? Tokens represent the amount of text the AI can "read" at once. More tokens = more market data can be analyzed in each query. For most users, the 8K model is more than sufficient. Use 32K or 128K if you enable candle history with many bars or use multiple indicators.

How to get your Moonshot API Key:

  1. Go to platform.moonshot.cn
  2. Create an account
  3. Navigate to API Keys
  4. Generate a new key
  5. Paste the key in your keys.txt on the moonshot: line
ℹ WebRequest URL
Add https://api.moonshot.cn to MT5 WebRequest settings.
Baichuan (百川智能)

Baichuan AI is a Chinese company specializing in high-quality language models with particular strength in Chinese language processing. If you prefer to receive your trading analysis in Chinese, or if you use Chinese-language custom prompts, Baichuan will deliver the most natural and accurate results. It also offers a 128K context variant for heavy data analysis.

ModelDescriptionBest For
Baichuan 4Most advanced modelMaximum quality analysis
Baichuan 3 TurboFast and efficientDaily use, automatic mode
Baichuan 3 Turbo 128KExtended context versionAnalysis with extensive historical data

How to get your Baichuan API Key:

  1. Go to platform.baichuan-ai.com
  2. Create an account
  3. Navigate to API Keys
  4. Generate a new key
  5. Paste the key in your keys.txt on the baichuan: line
ℹ WebRequest URL
Add https://api.baichuan-ai.com to MT5 WebRequest settings.
China Providers Comparison
FeatureQwenZhipu GLMMoonshotBaichuan
Free tierNoYes (Flash) — no depositNoNo
CostAffordableFree (no deposit) / AffordableAffordableModerate
Best modelQwen MaxGLM-4-PlusV1 128KBaichuan 4
Key strengthReasoningFree accessLarge contextChinese language
Backed byAlibabaTsinghua Univ.Moonshot AIBaichuan AI
VPN neededNoNoNoNo
Models available4433
★ Multi-AI with Chinese Providers
You can combine multiple Chinese AIs in the Multi-AI Consultation system. For example: Qwen + Zhipu + Moonshot with Majority decision. This gives you diversified analysis without relying on any Western service.

Multi-AI Consultation

Query multiple AIs simultaneously and combine their analysis

Have multiple AIs analyze the market at the same time and combine their opinions — like consulting several analysts before making a decision.

ParameterDefaultDescription
Enable Multiple AI ConsultationDisabledActivate multi-AI mode
Include OpenAI / Claude / Gemini / etc.VariousSelect which global AIs participate
Include Qwen / Zhipu / Moonshot / BaichuanDisabledInclude Chinese AIs in the consultation
Include OpenRouter #1 to #5#1-3: YesSelect which OpenRouter slots participate
Decision TypeMajorityHow the final decision is made
AI count for Majority Custom3How many AIs must agree (Custom mode)
Decision ModeHow It WorksExample (3 AIs)
MajorityMore than half must agree2 say BUY, 1 says SELL → BUY
Majority CustomExactly N AIs must agreeSet to 3: all must say BUY
UnanimousALL must agree, or no trade3/3 must say the same

General AI Settings & Custom Prompt

Temperature, timeout, language, and custom instructions
ParameterDefaultDescription
Temperature0.7AI creativity (0.0 = consistent, 1.0 = creative). Trading: 0.3-0.7
Maximum tokens in response1500Max AI response length
AI Status Check Frequency60 minVerify AI is responding
Timeout in milliseconds5000Wait time before canceling (increase to 10000-15000 for slow connections)
Response LanguageEnglish6 languages: EN, ES, FR, IT, PT, DE
Enable Custom Prompt from fileDisabledLoad custom instructions from text file
Custom Prompt filenamecustom_prompt.txtFile in MQL5/Files folder

Create custom_prompt.txt in MQL5/Files with your own trading instructions:

Only trade in the direction of the main trend.
Prioritize the London and New York sessions.
Use a minimum 1:2 risk/reward ratio.
Avoid trading during the Asian session.
Trade Filters

Economic News Filter

Pause trading around high-impact news events
ParameterDefaultDescription
Enable News FilterDisabledEnable/disable news filter
News Impact LevelHighHigh, Medium, Low, or combinations
Minutes before news30Block window before event
Minutes after news30Block window after event
Auto-update interval10Minutes between calendar refreshes

Example: High-impact event at 14:30 with 30-min buffers → no trading 14:00–15:00.

Fractal Multi-Timeframe Filter

Only trade when multiple timeframes confirm direction
ParameterDefaultDescription
Enable Fractal FilterDisabledActivate multi-TF analysis
Include M5/M15/M30/H1/H4/D1M15 & H1 OnWhich timeframes to check
Minimum confirmations2Timeframes that must agree (2-6)
Lookback periods10Bars to analyze
Minimum trend strength0.6Strength threshold (0.5-1.0)

Support/Resistance & Spread Filter

Filter by key levels and broker spread
S/R Filter
ParameterDefaultDescription
Enable S/R FilterDisabledActivate S/R analysis
Lookback periods50Bars to scan (20-200)
Minimum touches3Bounces to confirm level (2-5)
Touch tolerance0.1% of ATR tolerance
Spread Filter
ParameterDefaultDescription
Enable Spread FilterDisabledBlock when spread is too high
Maximum spread (pips)3.0Max allowed spread
Use average spreadEnabledAverage vs. instant
Average periods10Periods for averaging

Sideways Market Filter

Block trading when there is no clear trend (ADX-based)
ParameterDefaultDescription
Block Trading in Sideways MarketDisabledBlock when ADX below threshold
ADX Threshold20.0Below this = ranging market

Volatility System

Classify market conditions: LOW / NORMAL / HIGH
ParameterDefaultDescription
Use VolatilityEnabledActivate classification
Volatility Period14ATR period

Compares current TF volatility to daily ATR. Classification (LOW/NORMAL/HIGH) is sent to AI as context — does not block trades directly. 18 customizable thresholds per timeframe (M1-MN).

Trading Engine

Trading Panel — Auto Mode

4 trading modes, query frequency, and confidence threshold
ParameterDefaultDescription
Auto Mode EnableEnabledFully automatic trading
Trading ModeTimeTime / Breakout / Trend Follow / Pullback
Frequency (minutes)10Minutes between AI queries
Minimum Confidence (%)70AI must be this confident to trade

Modes: Time = query every X minutes. Breakout = query on range break. Trend Follow = query when trend confirmed. Pullback = query on pullback within trend.

Lot Management & SL/TP

Position sizing, stop loss, take profit, and risk-reward ratio
Lot Modes
ModeDescriptionExample
FixedSame lot every tradeAlways 0.01 lots
Percentage% of balance as risk2% of $10K = $200/trade
DynamicExact lots so SL = configured %50 pip SL at 2% risk → calculated
SL/TP Modes
ModeDescription
AI RecommendationAI chooses based on analysis
Fixed PipsFixed SL/TP in pips
Swing ATRCalculated from volatility
LearningAdaptive SL/TP from Learning System
DisabledNo SL or TP (not recommended)
Risk-Reward Ratio Override

With Use ratio system enabled and ratio 1:2, if AI sets SL to 30 pips, EA automatically adjusts TP to 60 pips.

Trailing Stop & Breakeven

Protect profits dynamically, virtual hidden SL/TP
ParameterDefaultDescription
Trailing ModeOffOff / Classic Pips / Percentage
Trailing Start50Profit pips to activate
Trailing Stop30Distance from price
Trailing Step20Min movement to update
Breakeven EnableDisabledMove SL to entry after X points profit
Breakeven Points30Points needed to trigger
Hidden SL/TP EnableDisabledVirtual SL/TP invisible to broker

Cooldown & TP-to-Breakeven

Wait after trade closes; emergency TP adjustment
ParameterDefaultDescription
Enable CooldownEnabledWait after trade closes
Cooldown minutes1Wait duration (1-180)
After SL / After TPBoth OnWhen to apply
Per symbolEnabledPer pair vs global
Enable TP to BEDisabledMove TP to entry when trade goes against you
Max drawdown before BE20 pips / 2%Trigger threshold

Overtrade Control

Prevent duplicate and excessive trades
ParameterDefaultDescription
Overtrade ProtectionEnabledBlock multiple trades on same pair
Min time between trades60 secMinimum seconds gap
Min pips distance50Minimum pip distance between entries

Trading Hours Control

Define trading windows per day of the week
ParameterDefaultDescription
Enable Trading HoursDisabledActivate schedule
Enable Weekend TradingDisabledAllow Sat/Sun
Close at end timeDisabledAuto-close positions
Enable Friday closingEnabledClose Friday positions

Per-day settings: Start Hour/Minute, End Hour/Minute (server time, 0-23). Supports overnight windows (e.g., 22:00–06:00).

Magic Numbers & Comments

Unique trade identifiers
ParameterDefaultDescription
Magic Manual / Auto123456 / 789012Unique IDs per trade type
Comment Manual / AutoMANUAL / AUTOLabels in trade history
⚠ Multiple Charts
Each ARIA instance must have unique magic numbers.
Indicators

Technical Indicators

RSI, MACD, Bollinger, Fibonacci, S/R — all fed to the AI
IndicatorOnParametersPurpose
RSIYes14Overbought/oversold
ATRYesVolatility
MA 200 / MA 50YesLong/medium trend
MACDYes12/26/9Momentum & crossovers
StochasticYes5/3/3OB/OS + momentum
Bollinger BandsYes20 / 2.0Volatility bands
FibonacciYes20Retracement levels
Support/ResistanceYes20Key bounce levels

Indicator values are sent to the AI as data — they don't generate signals alone.

Advanced V5 Indicators

Ichimoku, CCI, Heiken Ashi, EMA Slope
IndicatorOnParametersBest For
Ichimoku CloudNo9/26/52Trend trading H1-D1
CCINo14Price extremes
Heiken AshiNoTrend confirmation
EMA SlopeNo20/3Movement strength

Candle History & Additional Timeframe

Send OHLCV data and multi-TF context to AI
ParameterDefaultDescription
Include candle historyDisabledSend OHLCV CSV data to AI
Candle bars200How many candles
Additional TimeframeDisabledAnalyze a second TF
Selected TimeframeH1M1 through MN1
Risk & Protection

Risk Sentinel

Multi-layered circuit breaker: GREEN → YELLOW → ORANGE → RED
ParameterDefaultDescription
Enable Risk SentinelEnabledActivate guardian
Max Open Positions5Total open trades limit
Max Per Symbol2Per-pair limit
Currency Exposure CheckEnabledMonitor per-currency lots
Max Net Lots/Currency1.0Max in any single currency
Max Total Lots2.0All positions combined
CB Warning / Throttle / Kill DD%3% / 4% / 5%Circuit breaker thresholds
CB Kill Consecutive Losses6Losses in a row to trigger RED
Trailing DD Limit / Warning6% / 4%FTMO-style from equity peak
Daily / Weekly Loss Limit5% / 8%Period-based limits
Max Open Risk %5.0Total equity at risk
StateAction
GREENNormal trading, full lot
YELLOWWarning displayed, trading continues
ORANGELot reduced to 50% (configurable)
REDAll trading halted

Worst state always wins. Currency exposure is decomposed per pair (BUY EURUSD = LONG EUR, SHORT USD).

Daily Profit & Loss Limits

Auto-stop at daily targets or limits
ParameterDefaultDescription
Enable Profit LimitDisabledCap daily profit (Fixed $ or %)
Profit Limit Amount1000Target value
Enable Loss LimitDisabledCap daily loss (Fixed $ or %)
Loss Limit Amount500Limit value

Prop Firm Mode

FTMO, MyForexFunds, TFT rule enforcement
ParameterDefaultDescription
Enable Prop FirmDisabledActivate all protections
Max Daily Loss %5.0Daily loss cap
Max Total DD %10.0Total drawdown cap
Close Before WeekendDisabledFriday close
Block HedgingDisabledNo simultaneous BUY+SELL
Max Trades Per Day0 (off)Daily count limit
Max Single Trade %0 (off)Consistency rule
RuleFTMOMFFTFT
Daily Loss5%5%3%
Total DD10%12%6%
Weekend CloseNoYesYes
Block HedgingNoYesYes
Optimization & Analysis

Session Optimizer

A-F grading of your best and worst trading hours

Analyzes your trade history to grade each session (Asian/London/NY/Overlap/Off-Hours) and each hour from A to F. Grade D reduces lot to 75%, grade F to 50% or blocks trading.

ParameterDefaultDescription
Enable Session OptimizerDisabledActivate analysis
Min Trades per Session/Hour5 / 3Minimum data to evaluate
Cache Duration (min)30Report refresh interval

Config Auto-Optimizer

18-point analysis with automatic parameter tuning and rollback
ParameterDefaultDescription
Enable OptimizerDisabledActivate auto-optimization
Optimizer ModeSuggestSuggest / Auto Safe / Auto Full
Trader ProfileBalancedBalanced / Aggressive / Conservative / Prop Firm
Use AI AnalysisDisabledAI-powered insights
Rollback After24 hoursAuto-revert if change worsens results

Analyzes 18 aspects: best/worst hours, days, directions, R:R ratio, confidence level, trade duration, hour+day combos, losing streaks, and more.

Strategy Tester (Backtester)

Walk-Forward, Monte Carlo, Prop Firm compliance checks

Test strategies on historical data without AI or real money. Uses indicator-based scoring.

ParameterDefaultDescription
Min Signal Score3Score 1-10 to open trade
ATR SL/TP Multipliers1.5 / 2.5Volatility-based stops
Walk-ForwardEnabled (70/30)In-sample/out-of-sample split
Monte Carlo1000 iterationsRandomized trade order simulation
Prop Firm CheckEnabledVerify funded account compliance

Metrics: Sharpe ratio, Sortino ratio, Calmar ratio, recovery factor, P95/P99 drawdown.

Trade Memory System

Learns from past trades to auto-block bad patterns

Records up to 2,000 trades per symbol/TF. Tracks win rate by hour, day, direction, and session. Auto-actions:

  • Block underperforming hours and directions
  • Reduce lot after 3+ consecutive losses
  • Halt trading after 5+ consecutive losses

Data saved as CSV in MQL5/Files/AriaMemory/ (Excel-compatible).

Learning System

Adaptive SL/TP based on 500-bar ATR volatility analysis

Selectable as SL/TP mode "Learning". Analyzes 500 bars of ATR data to create 4 volatility thresholds (50th/69th/84th/93rd percentiles). Low volatility = tighter stops; high volatility = wider stops. Thresholds auto-calculated or manually set.

Integration

Telegram Bot

Real-time notifications and remote control
ParameterDefaultDescription
Enable TelegramDisabledActivate bot
Bot Token / Chat IDEmptyFrom @BotFather / @userinfobot
Notification LevelAllAll / Trades / Alerts / Critical
Remote CommandsEnabledControl EA from Telegram
Setup (3 steps)
  1. @BotFather in Telegram → /newbot → copy token
  2. @userinfobot/start → copy your numeric Chat ID
  3. MT5: Tools → Options → Expert Advisors → add https://api.telegram.org
Remote Commands
CommandDescription
/status (/s)Balance, equity, positions
/positions (/p)Detailed open positions with P/L
/riskRisk Sentinel + Prop Firm status
/report (/r)Trigger daily summary
/pause / /resumePause/resume auto trading
/help (/h)All available commands
✖ Without WebRequest URL, the bot will NOT work
This is the #1 setup error. Always add https://api.telegram.org.

Visual Customization

Chart colors, grid, volume — purely aesthetic

9 customizable parameters: background, foreground, bull/bear candle colors, bar colors, grid toggle, volume toggle and color. Default: professional dark blue theme. Does not affect trading or AI.

What's New
📋

ARIA CONNECTOR EA v5.20 — Update Notes

New free AI providers, verified OpenRouter models, and bug fixes
🔄 OpenRouter Free Models — Important Changes

After extensive testing with our new ARIA Model Tester, we identified significant instability issues with OpenRouter's free models. Out of 29 available free models, only 7 return correctly formatted responses that ARIA can interpret.

  • Many free models were discontinued or removed by their providers without notice
  • Several models return incomplete or malformed responses (missing ADVICE, CONFIDENCE, SL or TP)
  • Some models are frequently rate-limited (HTTP 429), making them unreliable for real-time trading signals
  • Models like nemotron-3-super-120b only scored 2/5 on our quality tests, while others scored 0/5

We ran every free model through an automated quality test, evaluating them on 5 criteria: ADVICE extraction, CONFIDENCE, STOP LOSS, TAKE PROFIT, and CONCLUSION. Only models scoring 5/5 remain in the list.

✅ New Free AI Providers: Groq & Cerebras

Two new free AI providers added that connect directly to their APIs — no OpenRouter needed. See the Groq, Cerebras & Gemini section for full details, available models, rate limits, and setup instructions.

✅ Google Gemini as Free Provider

Gemini continues to be one of the best free options with generous limits and high-quality responses. Recommended free models: Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini 1.5 Flash, and Gemini 3 Flash.

🔧 Other Improvements in v5.20
  • Fixed OpenRouter response parser — now correctly extracts ADVICE and CONFIDENCE from any model, even those using markdown format
  • Fixed SL/TP display — now shows only the price value instead of the full formula
  • Groq and Cerebras now appear in the AI Status Panel with real-time connection status
  • NVIDIA NIM provider removed (unstable API, limited model access)
  • Updated WebRequest URLs — remember to add the new URLs in Tools → Options → Expert Advisors:
    • https://api.groq.com
    • https://api.cerebras.ai
Reference

FAQ & Troubleshooting

Common questions and solutions
ProblemSolution
EA doesn't tradeCheck AutoTrading enabled, API key valid, WebRequest URLs added, confidence not too high
AI returns errorsVerify API key. Increase timeout to 10000-15000ms. Check WebRequest URL is added.
Telegram silentAdd https://api.telegram.org to WebRequest. Verify token and Chat ID.
China: AI not connectingAdd the correct Chinese provider URL to WebRequest. Ensure you're using Qwen/Zhipu/Moonshot/Baichuan, not Western providers.
Multiple EAs conflictingEach instance needs unique magic numbers.
QuestionAnswer
Best AI for beginners?OpenRouter with free models (global) or Zhipu GLM-4-Flash (China)
Multiple pairs?Yes, one EA per chart with different magic numbers
Best timeframe?M15 and H1 are most popular
VPS recommended?Yes, for 24/7 uninterrupted operation
Can I backtest?Yes, built-in Strategy Tester (indicators only, no AI during backtest)
Minimum balance?As low as $100 with 0.01 lots